The Zodiac Killer was a serial killer in the late 60's and early 70's. The twist is, he would frequently taunt the local press with cryptic letters. Four of these letters were actually encoded, but only one has been cracked to date.

I'm doubtful that the Zodiac Killer was a master cryptographer, at least not to the degree that one would be considered today. So why haven't his remaining three letters been cracked yet? Is this a technical problem, or is it just not being worked on? Are the letters available to the public for someone to try their hand with?


2 Answers 2


The Zodiac killer ciphers are an interesting case. As there were four ciphers sent to the local papers, I will address each in turn. They do share some common traits however.

  • They are each their own cipher, so the 'solution' used for cipher 408 cannot be applied to the other messages.
  • Each message has a unique character count.
  • The Zodiac Killer sent these ciphers to newspapers with accompanying plain text messages threatening to commit additional murders (including killing school children) were the ciphers not printed in the papers.


The message containing this cipher was sent to the authorities postmarked April 20, 1970.

This cipher was prefaced with the plain text "my name is". While the cipher contains 13 symbols, 8 of those symbols are unique. Attempts have been made to find a solution to this cipher using modern computing techniques, however the message is just too short and the possibilities are just to many to determine what 'solutions' are valid candidate solutions.


The message containing this cipher was sent to authorities postmarked June 26, 1970.

This cipher accompanied a plain text message claiming that when decoded this cipher would lead to the location of a bomb he had buried and set to go off at a later date. The cipher was never decoded and the alleged bomb never found.

The cipher had 32 symbols, 29 of which were unique - leaving pretty much nothing to analyze, essentially making it a one time pad cipher.


The messages containing this cipher were sent to the Vallejo Times Herald, the San Francisco Chronicle, and The San Francisco Examiner postmarked August 1, 1969. Each letter included one-third of the 408 symbol cipher.

On August 8, 1969, high school teach Donald Harden and his wife Bettye of Salinas, California, solved it, however the solution did not work for the last 18 symbols of the message. The cipher was a homophonic substitution cipher, but not a simple one. Rather than using 1 symbol = 1 letter substitution, The Zodiac Killer assigned multiple symbols to each letter. It is theorized that the last 18 letters of the cipher are filler symbols added to ensure that the three papers received equal portions of cipher text. There have been attempts to find a meaning for this last potion of the cipher "EBEORIETEMETHHPITI" via anagram, however there are 740 billion ways to anagram 18 letters and nothing meaningful has come of these attempts.

It's notable however that the rotation of symbols for each letter was consistent across the cipher until the last 18 letter, lending credence to the filler explanation. It's also interesting to note that The Zodiac Killer appears to have been a bad speller, or perhaps they struggled to follow their own code keys as there are spelling errors throughout the text. It's interesting however that these errors always seem to be made where the incorrect symbol resemble the intended symbol. (e.g. 'MOAT' instead of 'MOST' where the symbol for 'A' is a filled in triangle and the symbol for 'S' is a triangle with a dot in it.


The message containing this cipher was sent to authorities postmarked November 8, 1969.

There are a number notable differences between the 340 cipher and the 408 cipher which make 340 more difficult to 'solve'. The first difference is that it is 68 characters shorter, giving a substantially shorter message to analyze. 7 symbols which appeared in 408 were removed from 340 and 16 entirely new symbols were added in. These changes in addition to the shorter message have posed a real problem for analysis efforts.

Methods of analysis:

  • Symbol repetition by row and column - 340 has many rows but no columns without repeating symbols, which supports that the text is meant to be read horizontally. There is also a pattern, where the first three rows contain no repeated symbols, and then exactly at the half way point, another three rows which contain no repeating symbols. This has spawned a theory that the original cipher was created twice as wide, then cut in half vertically to put one side above/below the other as a means of obstruction.

  • Repeating Bigram analysis - when comparing 408 to 340, it is notable that 408 has 62 bigrams where as 340 only has 25. This is interesting because if you look at bigrams in randomly scrambled cipher text, you will normally get around 20 in this sized text. This would not support a horizontal text pattern. Additionally, if you cut the text in half - the first half has significantly more bigrams than the second half. This would indicate that something in the cipher is disrupting the naturally occurring bigrams.

  • Repeating Trigram analysis - when comparing 408 to 340, it is notable that 408 has 11 trigrams where as 340 only has 2. Whatever is disturbing the bigram count does not appear to be disturbing the trigram count. This favors the theory of a horizontal reading direction.

  • Bigram distance analysis - This type of analysis looks at the possibility of symbol transposition. This analysis as been inconclusive and no conclusions have been able to be drawn.

  • Homophonic substitution analysis - Strongest example (symbols L/M) repeat 7 times.

  • odd/even analysis - Very wide spread of bigram counts. Less than 1% of randomized trials show this even of a bigram spread.

Notable Oddities

  • Two pivot pairs appear in the cipher text. The chances of this happening naturally Once in this size of text is 1 in 50,000. This may or may not be significant.

  • Two strange 'box corner' patterns appear in the text between the 'O' and 'C' symbols.

  • "ZODAIK" seems to appear in the cipher at the bottom left of the last line of text, if you replace a solid triangle symbol with a 'D'

My Conclusion

To answer your questions specifically;

Why haven't his remaining three letters been cracked yet?

13 is too short. 32 is essentially a one time pad. 340 is short, complicated and nigh-on nonsensical.

Is this a technical problem, or is it just not being worked on?

It's a technical problem. without patterns, no amount of pattern analysis is going to make a difference.

Are the letters available to the public for someone to try their hand with?

Yes. There are multiple forums dedicated to this task.

The 340 cipher is still being actively pursued by a dedicated community. There are several forums (http://zodiackillersite.com, http://zodiackillerfacts.com, http://zodiackiller.com) focused on finding a solution. The shortness of the cipher and the number of symbols pose a real challenge to investigators. It would be interesting to see if an AI like Watson or Deep Blue would be able to make any headway on this, however at this point, to some degree its academic. No solution to this cipher will be able to bring any peace to those affected by the murders.

If you are truly interested in this topic, I would HIGHLY recommend you listen to the video I've listed in my sources and visit the forums. The people who are working on this will be able to go into FAR more depth than I'm able to.



Zodiac Killer Ciphers

Youtube - David Oranchak

  • "these errors always seem to be made where the incorrect symbol resemble the intended symbol" -- that's really scary, because it suggests that the message was encrypted by one person and then transcribed by another. May 8, 2016 at 19:59
  • It could also be that he wrote it down temporarily on a scratch pad as he was enciphering the text, and then copied the completed ciphertext to his letter. It'd be hard to believe he wrote the enciphered text in one go onto the note.
    – forest
    Nov 29, 2017 at 21:37
  • This was solved today! thehill.com/homenews/news/…
    – JesseTG
    Dec 12, 2020 at 5:14

You're assuming that they're actually encrypted. A lot of crazy people have written things that nobody understands. Just because the author thinks they're in code doesn't necessarily mean that the code can be reversed.

  • 7
    I agree with you but a small correction is needed. It should be 'encoded' instead of 'encrypted'. Encryption implies a presence of a key to decrypt the data, while encoded means there is no key, just steps. May 8, 2016 at 17:16
  • It might also have been a ruse: have folks focus on decoding strings of random letters, rather than on more concrete clues. Not to mention that if you can/have to accommodate misspellings then you can probably produce a whole bunch of seemingly valid phrases out of any similarly sized random string...
    – thkala
    May 8, 2016 at 17:33

Not the answer you're looking for? Browse other questions tagged or ask your own question.